Last week I launched a small new campaign for a client.
Following best practices I added two ads to the campaign and set them to “Rotate Indefinitely: Show lower performing ads more evenly with higher performing ads and do not optimise”.
There is a caveat below this option which states : “Not recommended for most advertisers” – but I like living on the edge.
I also like to give my target audience the benefit of the doubt that they might know better than I what they will respond to. This is the essence of rotating your ads. run a number of ads against one another, see which ones people respond to best – not just in terms of CTR, but also with respect to other measures of engagement and, of course, conversions.
In my experience I have often been surprised how ads which I had thought were terrible and should never have been allowed to see the light of day, perform wonderfully and without shame once launched on an unsuspecting target group.
I checked back a couple of days later to see that these ads, set to rotate evenly were doing anything but.
I posted a quick, admittedly slightly tongue in cheek, image to twitter:
Twitter, in its wonderful way, responded :
and any number of possible explanations started to appear :
The settings weren’t changed, but @bloomarty ‘s comment would clearly have explained the behaviour. But what @philip_a had experienced and what @MaltzPPC suggested were on the right track.
Another of the #ppcchat members, @Realicity suggested segmenting the keywords by ad position.. and this shed some further light on the topic :
The segmentation showed that, whilst for some of the keywords the positions were the same or similar,for most, one of the ads was in a significantly worse position on average :
and then @AdWords responded :
Quality Score is a big factor.
All along my assumptions (and I now realise that they were assumptions) about “rotate evenly” meant were misguided.
I had, mistakenly (apparently) assumed that “rotate evenly” meant that me ads would be, well, rotated evenly. But this is not the case.
Further discussion with Eamon at @AdWords – for which I am very grateful – cleared things up further :
The bids are normally set at the ad group level, extensions at the campaign level, but the ad copy is a factor in Quality Score – so the keywords are quite likely to have different QS depending upon which ad is considered for the impression.
But here’s the thing – for me – this is not rotate evenly. Surely the whole point of rotating your ads is to see how well they perform compared to the other ads in the ad group. We know that one ad will give a keyword a higher QS – what we don’t know is which ad will resonate most effectively with our target audience and lead to more conversions, sales or engagement. It may be the one which has the poorest Ad Rank.
I see no way around this.
What I do see, however, is that I perhaps need to rethink what exactly I am testing when I run an “evenly rotated” set of ads.
Finally, there is another lesson to be learned in the reply from AdWords – if you run your ads to rotate indefinitely, and do not change then for 90 days, then Google will revert your setting to optimise for clicks or conversions. Indefinitely = 90 days. not such a biggie – if you don’t change your ads in 90 days then you deserve to have someone come in and clean up your mess, but still….